RealClimate logo

Fun with correlations!

Filed under: — gavin @ 9 May 2007 - (Türkçe)

We are forever being bombarded with apparently incredible correlations of various solar indices and climate. A number of them came up in the excoriable TGGWS mockumentary last month where they were mysteriously ‘improved’ in a number of underhand ways. But even without those improvements (which variously involved changing the axes, drawing in non-existent data, taking out data that would contradict the point etc.), the as-published correlations were superficially quite impressive. Why then are we not impressed?

To give you an idea, I’m going to go through the motions of constructing a new theory of political change using techniques that have been pioneered by a small subset of solar-climate researchers (references will of course be given). And to make it even more relevant, I’m going to take as my starting point research that Richard Lindzen has highlighted on his office door for many years:

That’s right. Forget the economy or the war(s), the fortunes of the Republican party in the US Senate are instead tied closely to the sunspot cycle.

“Oh yes”, the sceptics might say “but that’s just a couple of cycles and doesn’t use up-to-date numbers. What happens after 1986?”

Well, that is a little problematic, however, the good early correlation is obviously still important (r=0.52! 1960-1986) and so we should be able to refer to it over and over again without noting that it breaks down subsequently (cf. Svensmark, 2007 referring to Marsh and Svensmark (2000)). But more importantly, it just demonstrates that the theory needs a little adjustment.

Let’s look at the second half of the record. Well, there’s another strong correlation for that period as well (r=-0.63, 1988-2006). Only this time the correlation is inverted, but that shouldn’t be surprising to anyone – solar-senator effects are complicated!

If we now put it all together, we can see that there is a reasonable match over the whole period…. well, except that break in the period 1984 and 1988 and, unfortunately, last year’s elections didn’t fit the pattern either. But 1984-1988 was Ronald Reagan’s second term and clearly no theory of Republican senators can ignore that. We therefore propose that the ‘Ronald Reagan second term phase shift’ combined with the change of sign of the Hale solar magnetic cycle in 1986, obviously changed the dynamics. This kind of phase shift is frequently seen in solar studies (cf. Landscheidt and many others), where it is rarely taken as a sign that two time series with decadal spectral power are in fact completely independent. Finally, it is permissible to leave off the more recent data points (cf. TGGWS) for “graphical convenience”. So after just a little work, we have managed to rescue the original theory to match a much longer amount of data:

Some readers may scoff and suggest that in the absence of any mechanism, these powerful correlations are numerological artifacts arrived at using post hoc fallacious reasoning that have no predictive capability. That might appear to be a valid argument. However the ultimate test will of course be experimental. On the basis of these intriguing results, we propose exposing Republican senators to varying levels of cosmic rays in a basement and monitoring their electability. Any refusal by the funding agencies or ethical review panels to support this would simply be confirmation that the political science establishment are scared of what this research would imply for their so-called “consensus”.

Convincing, eh?

The data for sunspots and senators can, I’m sure, be manipulated even more effectively than I’ve done here. I’ve made no use of various lags or filters (which can be altered as you go along cf. Friis-Christensen and Lassen (1991)), or of partial detrending (cf. Marsh and Svensmark (2003)), or of splicing of unconnected data sets (cf. Svensmark and Friis-Christensen 1997, Nir Shaviv). More ideas could be taken from “New evidence for the Theory of the Stork” (Höfer et al, 2004)”. A special RealClimate commendation for anyone who can do better!

356 Responses to “Fun with correlations!”

  1. 1
    Charles Raguse says:

    A respectable linear relationship (rising over time) could be fitted to the data.

  2. 2
    Jose Larios says:

    he he he that`s great post
    good job

  3. 3
    Aaron Lewis says:

    I had always suspected that the Democrats and Republicans had undergone a role reversal on October 22, 1986. It is nice to see proof. I am sure that if you had a statistics package designed specifically for the social sciences you could tie that role reversal to a specific sun spot event (or lack there of.)

  4. 4

    Well done, Gavin. I halfway expect Dennis Wingo to not only agree with you for once but also to actually be happy with the results you’re getting.

  5. 5
    Timothy Chase says:


    I think you should know that contrarians are being especially easy on the climatologists at this point. With evolutionary biology, we have to deal with the young earth creationists, the old earth creationists, the special creationists (who accept the fossil record, but claim that each kind was created in the era in which it appears indendently of all the rest), then all the possible variations.

    For example, omphalosian creationists argue that the world is young but accept the existence of all the evidence that the world is old, arguing that the world was created only 10,000 yrs ago, including all of the evidence which gives it the appearance of being very ancient. Of course one could argue that it was really created only five minutes ago, including even our memories which give it the appearance of being older. But then one isn’t talking empirical science – one has left that behind quite some time ago in favor of a freshman philosophy course bullsession.

  6. 6
    Thom says:

    Carl Wunsch has published an email exchange with Durkin, the producer of that phoney documentary.

    Durkin does not respond well to critique. “You’re a big daft cock,” he wrote to journalists and scientists.

    Durkin also asks why the issues raised in his wonderful opus have not come up in the “hours and hours of shit programming on global warming shoved down our throats by the BBC?”

    “Go and fuck yourself,” he ends.

    Seems like a nice guy. I’m sure that’s why Wunsch is publishing the emails, since Durkin did him such a nice favor by selectively editing an interview.

  7. 7
    NU says:

    A nice job of explaining “correlation does not equal causation”. However, your point would be stronger if you had concluded with a comparison to the CO2-temperature correlation issue: why we have strong reasons to believe a causal link in that case, but not in the case of sunspots and temperature. The two cases are not the same, but it would behoove you to point out why they are not the same, lest the fence-sitters in your audience accuse you of hypocrisy, or conclude that there is equally no reason to believe that CO2 has influenced temperature.

  8. 8
    tamino says:

    I notice you “conveniently” left out the data before 1962. Close scrutiny of the entire time series will show that over the long term, Republican senate membership is much more strongly tied to the length of the solar cycle, all the way back to the election of President Lincoln in 1860. Your mistaken correlation is simply due to the “urban liberal bias” in the historical record.

    And well before that, I can show strong correlation between solar intensity and conservative-vs-liberal dominance in politics. Contrary to IPCC claims, I can prove that the “medieval conservative period” was in fact even more conservative than modern times. Using Beryllium isotope abundances in meteor fragments as a proxy for solar activity, and the consumption of cheap beer as a proxy for conservative control of politics, I have shown a near-perfect correlation all the way back to Julius Caesar’s triumph over the bleeding-heart liberal Gauls.

    Just because I don’t have any actual data, doesn’t mean you can discount my theory.

    I came, I saw, I correlated.

  9. 9
    Knut Witberg says:

    This is odd to read. It looks like statistics and correlation is all new to you guys. If you pick 2 time series randomly from many time series, there is ~ 5 % likelihood that the two time series will significantly correlate at the 5 % level. It is therefore the science also must show that it is likely that the correlation is scientifically meaningful and logical. But this problem is the same for all research – whether you are a sceptic or a believer is not relevant.

  10. 10
    Hank Roberts says:,9171,758405,00.html
    [the comma in the URL breaks the link with this forum’s software; cut and paste works]
    Monday, Nov. 22, 1937
    “… it means more when one scientist with impeccable credentials declares that sunspots may have a physiological and emotional influence on mankind than when a thousand astrologers and other cultists affirm flatly that they do. …”
    “… Harlan True Stetson … His colleagues have voted him an asterisk in American Men of Science for distinguished research.”

  11. 11
    Jim Manzi says:


    Great post, but don’t you think that this also makes the general point that in the absence of a ‘mirror Earth’ to act as a control group it is extremely challenging to conduct reliable attribution studies in climate science?

    It seems to me that one common feature of economics, political science and climate science that distinguishes these disciplines from, say, physical chemistry, is the difficulty in ever executing a sufficient number of replicable falsification tests for such a highly over-determined outcome when we have only one experimental instance (the Earth). As an example, suppose I decided to test the predictiveness of the Sunspot – Republican Senate model by evaluating its performance in 2008, and it “predicts” the actual result perfectly? Even if we dreamed up some a priori potential causal relationship, I don’t think you or I would conclude that we had a useful tool, because there are so many factors that drive any election result.

    In the same way, suppose a GCM perfectly predicts a climate result for the next several decades — how reliable a falsification test is that one result? In the case of GCMs, this problem is exacerbated by the fact that a single falsification test requires 30+ years, so it’s not clear we even time to do one before we have to make decisions about changes to our behavior.

    It seems to me that the physics of infrared absorption and redirection (i.e., the known causal link) + common sense are what make the compelling case that human activities are most likely driving some amount of warming.

    Jim Manzi

    [Response: Climate science by correlation is not in the least bit satisfying, and yet it gets cited frequently – particularly by contrarians. I agree with you that physical modelling is the much better way to go (obviously since I work in a GCM group). However, I disagree that this cannot be tested. The models that are built can be tested in dozens of ways to verify key feedbacks and mechanisms – response to volcanic eruptions, solar forcing, orbital forcing, ENSO variability, NAO or SAM responses, the last glacial maximum etc. And frankly, the projections from simulations done 20 years ago stand up very well to what actually happened. A single ‘validation’ is of course not sufficient, but the hundreds of validations that have been done start to add up. – gavin]

  12. 12
    Ike Solem says:

    Brilliant! Here’s more proof that CO2 emissions are not the cause of the observed temperature increase, and that in fact, we have nothing to worry about.

    Fact #1: World population growth rates correlate very well with the global temperature trend. There is a spike around 1940 (the war years, when people were more active) followed by a post-war lull (when people were less active) – which just supports the mechanistic explanation that’s it’s human body warmth that’s causing this trend.

    Fact #2: As the above population link shows, global population growth rates are slowing, and in fact there is a top-heavy population structure in most European countries. As Third World countries develop, they too will follow this trend. See US census predicts slowing growth rate… yes, that’s a slowing rate, not a decreasing population, but if you extrapolate that trend forward one must conclude that global population will begin to decline late in this century.

    Conclusion: we should expand fossil fuel use as fast as possible in order to bring the Third World up to the development level of the US and Europe, which will lead to a decrease in human population – and that means less body heat, and therefore, global cooling.


  13. 13
    SecularAnimist says:

    Timothy Chase wrote: “For example, omphalosian creationists argue that the world is young but accept the existence of all the evidence that the world is old, arguing that the world was created only 10,000 yrs ago, including all of the evidence which gives it the appearance of being very ancient. Of course one could argue that it was really created only five minutes ago, including even our memories which give it the appearance of being older.”

    Actually the entire universe, including our memories and everything else that makes the universe appear “old”, came into existence one nanosecond ago. And this happens continually. Each and every nanosecond, a new universe comes into existence which is then replaced by a new one the next nanosecond.

  14. 14
    Nick says:

    You can explain GW in the same way. It correlates quite nicely with the number of TVs in the world.

    However, your analysis is just stupid and you miss the point.

    There is a clear correlation with solar activity over the long term record.

    That is not explained by solar radiation.

    Then the AGW activist claim that the solar activity can’t be the cause.

    A clear error in logic. There is another mechanism that does explain the correlations. The science is missing a mechanism and the AGW advocates have got quite rightly hot under the collar about one of the proposals.

    The reason is that it blows out of the water the anthropogenic effect of GW.

  15. 15
    Thomas says:

    @Nick (Post#14)

    Was your post intended to be nonsensical?

  16. 16
    Margo says:

    You said this:
    “And frankly, the projections from simulations done 20 years ago stand up very well to what actually happened.”

    Could you provide pre-1987 citations for for the simulations you believe stand up well to what happened in the past 5 years? And could you provide pre-volcanic explosion citations that agreed well with what happened after said volcano exploded. And so on?

    Presumably, if these predictive simulations were published before the events occurred, and later validated against what happened, it should be possible to post the early publication dates.


    [Response: Hansen et al, 1988 (simulations done in 1986/87), Hansen et al, 1992. – gavin]

  17. 17
    Steve Latham says:

    How do you know that senators of a certain type aren’t affecting sunspot activity? It seems to me that you’re assuming sunspots to affect electability when maybe the effect works in the other direction.

  18. 18
    Steve Horstmeyer says:

    Having spent 30 years in the U.S. television industry as a meteorologist (something I at times am slow to admit for obvious reasons), I have encountered many people who “do not get it”. Happily most I have encountered will defer to someone with a base of knowledge to ensure the accuracy of the final product. The “it” some do not get is that attention to minute details is what often separates “good science” from “bad science”. In that vein “adventure science” often crosses the line.

    There are many examples of “production value” trumping “scientific accuracy”. Sadly the very definition of “creativity” is convoluted in the process. What is ultimately described as creative by writers and producers of a science program is just the opposite, not creative and just maybe the easy (read – lazy) way out.
    The truly creative TV writer/producer finds a way to grab the audience and keep its attention using the facts at hand but many ignore the facts or invent situations because they cannot adequately deal with what they perceive as mundane. Of course we cannot ignore blatant advocacy and willful distortions, a possibility in the case described above. Then again maybe that situation is just ignorance.

    Before I go on I must emphasize that a TV program cannot read like a peer reviewed journal article. No one will benefit if the bored audience yawns then changes the channel. Even PBS ( for those not familiar it is the non-commercial Public Broadcasting System funded by tax payer dollars in the U.S.) seems to be catching on. “Brains boiled and heads exploded,” is how they promoted an episode of “Secrets of the Dead” about the encounter of the residents of Herculaneum with a pyroclastic flow from Mt. Vesuvius. I watched and the program delivered an accurate and very interesting portrayal of the sciences involved. Though it accurately describes what happened, I still am uneasy with the “brains boiling and heads exploding” imagery used for promotional purposes.
    Who can forget the giant block of permafrost swinging beneath a soviet helicopter with the ancient tusks of a mammoth protruding against the Siberian sky in a Discovery Channel documentary. The problem is the tusks were recovered three years earlier and this one scene mislead the audience about the tedious work of excavating in a harsh environment. The tusks were attached by a TV crew for dramatic effect.

    Or the “educational documentary” in a local TV market on the plate subduction volcanoes surrounding the Pacific Ocean called “The Ring of Fire”. The problem is that the show was video taped in Hawaii and to the amazement and disgust of geologists in that market no distinction was made between “hot spot” volcanoes like Hawaii and the volcanoes that result from the subduction of continental plates like those found in the Andes, the Aleutian Islands, Japan and Indonesia. I suppose Hawaii was an easier destination to deal with. The uninformed audience was mislead.

    My experience has shown that there is a dual problem in many lower budget TV productions: 1) Too many in the TV industry lack a broad base of knowledge and when it comes to science they are stumbling around in a wilderness bare foot and blind folded. 2) Without the economic resources to hire “experts” to guide the course of the program’s science writers and producers often opt for conventional wisdom, which of course is not always correct.

  19. 19
    Jacob Tanenbaum says:

    I like this theory. At least it seems more plausible than the idea that the majority of voters actually chose some of these guys. But forget about manipulating the data, what’s really important now is figuring out a way to adjust the number of sunspots so the Democrats win in 2008.

  20. 20
    Eli Rabett says:

    Personally I prefer the inverse pirate theory of global warming, showing that the lower the number of pirates the higher the global temperature

  21. 21
    Bruce Hall says:

    I loved your correlations and I loved the explanation in 2004 that even though the first 800 years of warming in a 5000 year trend were not caused by CO2, that didn’t mean that the other 4200 years couldn’t be.

    By that logic, I can say that just because some very wealth individual didn’t give me money in my first 6 decades they won’t from now on.

    Sorry, that a logical non-sequitur. Saying that because there wasn’t the proposed correlation in the past doesn’t preclude a future relationship, is a thinly veiled attempt to imply that there is/will be the proposed correlation (and reverse causation) in the future.

    Doesn’t work that way. Dung does not cause full stomachs in cows either.

  22. 22
    Wolfgang Flamme says:

    1987,1988… hey, this isn’t a phase shift, it’s just that global sulphur emissions started to decline since then.

    We all know this changes a lot because our level of scientific understandig with respect to sulphur emissions is rather low.

  23. 23
    Gareth says:

    Gavin, there’s no such word as excoriable. I think you mean execrable – which fits perfectly. On the other hand, you could excoriate the programme, perhaps making it “excoriable”… but I think we’d better leave constructing new words to the psychosolar connectionists.


    A Pedant.

    [Response: Hmmm… excoriate (def. #2) means to censor strongly; to denounce. Excoriable means capable of excoriation, and so implicity means ‘capable of being denounced’. I’d be happy with execrable as well of course…. – gavin]

  24. 24
    John L. McCormick says:

    RE # 20, Eli,

    I find your source for the higher priates and global warming is hardly credible. It carries less scientific rigor than that of a theory I found on another blog which attributes increasing SST with the repeatedly observed setting of the sun into the Pacific Ocean.

  25. 25
    Nick says:


    Not in the least bit nonsensical.

    The arbiter in cases such as GW has to be statistical.

    Climate change is not weather, it is long term observation and statistical analysis. You will no doubt agree that anyone who claims that a hot month in a particular local is proof of GW is barmy. The reason is that it is statistical proof that matters.

    In the case of GW, even the IPCC admits that there might (only might) be proof on the 50 year average. This is a very weak standard.

    Statistical methods in this case, because the evidence being claimed is statistical, is the way to go.

    Now the original article is clearly a spoof as you know. However, within it is the nub of the argument. The writer believes in AGW. They have attempted to show that correlation is not causation. In this case because there isn’t a physical explaination behind the causation.

    However they have missed the point. If there is correlation, it doesn’t mean that one particular causation explains all the correlation.

    i.e. To say that solar radiation explains only a small percentage of the correlation, doesn’t mean that solar effects are not present. It could be random chance (unlikely in this case) or more likely, that a particular physical explaination has been missed.

    Think about the statistical tests of this part.


  26. 26
    Jim Eager says:

    Re:21 “By that logic, I can say that just because some very wealth individual didn’t give me money in my first 6 decades they won’t from now on.”

    That’s not logic, it’s rhetorical analogy. And a pretty stupid one at that.

  27. 27
    DocMartyn says:

    I had a look at Hansen et al, 1988, how much of the temperature change over the last 20 years has been due to the present getting warmer and how much to the past getting colder?

  28. 28
    Walter Manny says:

    I had been told by a colleague that “RealClimate” was a good source for disinterested information on clmiate change. Not the right day to check in, I guess, but “Fun with Correlations” is a disqualifying front page article in that regard. I’ll keep searching, but have fun with your site, anyway!

  29. 29
    interested observer says:

    This post really is a poor effort at humor. Not that long ago you lost a debate because you were unable to present arguments sufficient to convince an audience of the merits of a position which, I suggest, you consider unassailable. Now we see this rather silly post. Try to stick to science rather than bad attempts at statistical satire. Otherwise you run the risk that readers will agree with poster #9’s comment that “This is odd to read. It looks like statistics and correlation is (sic) all new to you guys”.

  30. 30
    James Killen says:

    Re: #20
    Eli, the problem with the inverse pirate theory is that there is never any evidence presented that the number of pirates globally has actually decreased, in fact the opposite seems to be the case. Putting aside software piracy, actual maritime piracy may have decreased in the Caribbean and the Americas generally, but it flourishes elsewhere in the world, most notably in the Malaca Strait. To quote from Wikipedia:
    Piracy in the Strait has risen in recent years. There were about 25 attacks on vessels in 1994, 220 in 2000, and just over 150 in 2003 (one-third of the global total).

    Indeed according to one estimate “The total damage caused by piracy-due to losses of ships and cargo and to rising insurance costs-now amounts to $16 billion per year.” <;

    So much for the inverse pirate theory. I hasten to add that this in no way undermines FSM theory in general.

  31. 31
    Ike Solem says:

    #25 Nick, I still can’t tell if you’re joking or not. Are you trying to make fun of statistical analysis of correlations, i.e. r-factors?

    Assuming for a second that you’re being serious, an increased greenhouse effect due to increased greenhouse gases should cause cooling of the stratosphere…but warming due to increased solar output should cause warming of the stratosphere… or am I missing something? I feel like I’m falling for a joke, though.

  32. 32
    Eli Rabett says:

    Doc Martyn puts the steel toed boot in his mouth. Since, as far as we know the past is not currently changing and we have no good observations of the future, one can only compare predictions made in the past with what happened until today. The fit is quite good, excellent in fact, but Doc appears to have some unspoken problem with that.

    As to Gavin’s experience on the West Side, this little jibe is exactly what he needed. Something to illustrate, in an amusing way the falacies he was bombarded with.

  33. 33
    William Astley says:

    Is the purposed of this forum to discuss climate change? The 20th century warming was 0.6C in a 100 years. The 8200 yr cooling event was a 2C to 3C cooling in 100 years. Why did the Pacific Ocean cool 3C during the 8200 kyr event?

    The paleoclimatic record has a series of warm events followed by cold events. I will take bets that a cold event will follow the 20th century warm event.

    The 8200-year Climate Event

    I am curious how an abrupt stoppage or perturbation of the North Atlantic thermohaline circulation (THC) �Could have caused an:

    “Abrupt tropical cooling 8,200 years ago”, (3C cooling in 100 years) in Alor Indonesia, which is in the Pacific Ocean not the Atlantic Ocean, 3C cooling in 100 years.

    From M.K. Gagan et al.’s paper:

    “We drilled a sequence of exceptionally large, well-preserved Porites corals within an uplifted palaeo-reef in Alor, Indonesia, with Th-230 ages spanning the period 8400 to 7600 calendar years before present (Figure 2). The corals lie within the Western Pacific Warm Pool, which at present has the highest mean annual temperature in the world’s ocean. Measurements of coral Sr/Ca and oxygen 18 isotopes at 5-year sampling increments for five of the fossil corals (310 annual growth increments) have yielded a semi-continuous record spanning the 8.2 ka event. The measurements (Figure 2) show that sea-surface temperatures were essentially the same as today from 8400 to 8100 years ago, followed by an abrupt 3C cooling over a period of 100 years, reaching a minimum 8000 years ago. The cooling calculated from coral oxygen 18 isotopes is similar to that derived from Sr/Ca. The exact timing of the termination of the cooling event is not yet known, but a coral dated as 7600 years shows sea-surface temperatures similar to those of today.”

  34. 34
    Jim Roland says:

    I’m sure RC already know this, but now even the solar variability theorists have complained about TGGWS, for embellishing the solar variability data and unduly ruling out any AGW contribution:

  35. 35
    William Astley says:

    There does appear to be some strange correlation of solar activity and 20th century warming. Is this only a coincidence? Note the period of high solar activity is over. Will there be coincidental cooling also? Anyone one to make a bet?

    2005 paper by Georgieva, Bianchi, & Kirov �Once again about global warming and solar activity�

    CMEs, however, are not the only source of high speed solar wind. Early in the 20th century it was noticed that many geomagnetic storms occur without any visible solar disturbance. Such storms tend to recur every 27 days – the period of solar rotation, therefore they originate from long-living regions on the Sun which come back into geoeffective position rotation after rotation. Only when X-rays telescopes were flown above the atmosphere, it was found out that are large regions of open magnetic field geometry, and sources of high speed solar wind. They are now known as Coronal Holes (CHs) because, due to their lower density and temperature compared to the surrounding corona, they look darker in X-rays.

    In Figure 6 the long-term variations in global temperature are compared to the long-term variations in geomagnetic activity as expressed by the ak-index (Nevanlinna and Kataja 2003). The correlation between the two quantities is 0.85 with p<0.01 for the whole period studied. It could therefore be concluded that both the decreasing correlation between sunspot number and geomagnetic activity, and the deviation of the global temperature long-term trend from solar activity as expressed by sunspot index are due to the increased number of high-speed streams of solar wind on the declining phase and in the minimum of sunspot cycle in the last decades.

  36. 36
    Timothy Chase says:

    Nick (#25), you are being disingenuous.

    No one in their right mind claims that the solar forcing plays no role. In fact the IPCC views the variation in solar behavior as having played a substantial role, perhaps even the majority role in forcing climate changes during the earlier half of the twentieth century. As such, you are being misleading when you state, “However they have missed the point. If there is correlation, it doesn’t mean that one particular causation explains all the correlation.”

    However, clearly carbon dioxide has played a substantial role over the past 500,000 years – and we have a considerable amount of evidence for that. As such, you are being misleading when you state, “In the case of GW, even the IPCC admits that there might (only might) be proof on the 50 year average. This is a very weak standard.”


    What we have is evidence from a vast number of independent lines of investigation all pointing to the same conclusion: that anthropogenic forcing is resulting in a dangerous rate of climate change. We can measure the level of carbon dioxide being emitted. We can identify its manmade origin by tracking the isotopes. We can identify the highly robust correlation between carbon dioxide and global temperatures which have existed for at least the past 500,000 years.

    Moreover, we know how the physical principles by which global temperatures and levels of carbon dioxide are related. Increased temperatures in the past (e.g., due to changes in the earth’s orbit) have resulted in the earth’s climate entering a state of non-equilibrium, where more carbon dioxide is emitted than naturally sequestered, leading to a positive feedback loop. For example, with higher temperatures, the ocean’s capacity to sequester the carbon dioxide that it is already saturated with is lowered, resulting in the emission of carbon dioxide into the atmosphere.

    However, as carbon dioxide absorbs wide bands of infrared light (i.e., the earth’s blackbody radiation), it absorbs then re-emits thermal energy, some of which is reabsorbed then re-emitted by the surface. As a consequence, there exists positive feedback between the thermal energy emitted by the ground and the thermal energy emitted by the atmosphere – but where some of the thermal energy is lost to space in each successive round – necessarily avoiding a vicious “run-away” effect.

    For the earth-atmosphere system to reach a new equilibrium, the amount of energy going into the system has to equal the amount of energy leaving the system, but with more carbon dioxide in the atmosphere slowing the dissipation of such energy from the earth, this can only be achieved at a higher temperature where the thermal energy radiated by the system as a whole into space is once again raised so that it equals the thermal energy being received from the sun.

    If one increases the temperature of of the earth by increasing the the amount of energy it receives from the sun, this raises the level of carbon dioxide. But likewise, if one raises the level of carbon dioxide, this increases the temperature. Positive feedback – like the “thermal flux” feedback between the earth and its atmosphere that keeps the average temperature of the earth above freezing. It effectively comes to an end when a new equilibrium is reached. Simple, demonstrable physical principles.


    You know that hot bodies emit radiation: you have seen embers and hot metals glow with heat. You know that infrared radiation exists: you have seen infrared night vision – and you know that bees are able to navigate by the sun on an overcast day. You know that materials can be opaque to a given part of the spectrum: you have seen colored filters. You know that when matter absorbs radiation, it heats up: no doubt you have seen and felt this in the case of asphalt on a hot summer day. You know that we can measure different parts of the electromagnetic spectrum, from gamma rays to longwave radio frequencies. You know that the sky is blue because it is opaque to and scatters blue light.

    Each and every step is something which you can no doubt understand. Undoubtedly you realize that empirical science has a genuinely scientific grasp of the principles involved – which goes well beyond what either you or I could know without a great deal of study. Yet you fall back upon upon freshman philosophy in opposition to what is known by means of modern empirical science. You latch on to Hume’s critique of causality as if to claim that no matter how many times we see a hammer fall to the ground when released, we have no more reason to think that it will fall if released again than if we were as innocent of experience with hammers as a newborn babe. You avoid what should now be as clear as day, like a modern-day Descartes staring at his hand with limitless doubt.


  37. 37
    Chuck Booth says:

    Re # 9, 29 [It looks like statistics and correlation is all new to you guys.]

    Let’s see what qualifications these RC contributors have:

    Schmidt: BA (Hons) in Mathematics from Oxford University, a PhD in Applied Mathematics from University College London

    Mann: undergraduate degrees in Physics and Applied Math from the University of California at Berkeley, an M.S. degree in Physics from Yale University, and a Ph.D. in Geology & Geophysics from Yale

    Steig: BA from Hampshire College at Amherst, MA, and M.S. and PhDs in Geological Sciences at the University of Washington

    Connelley: B.A. in maths from St Edmund Hall, Oxford; doctorate in Numerical Analysis at Oxford

    Bradley: Director of the Climate System Research Center at the University of Massachusetts, Amherst and a University Distinguished Professor in the Department of Geosciences. B.Sc. 1969 University of Southampton, England; Ph.D. 1974, M.A., 1971 University of Colorado, Boulder; D.Sc 2003 University of Southampton

    and so on. Nope, no evidence I can see that anyone here knows beans about statistics.

    Re# 18 A bit off-topic, but years ago I knew a grad student who, while at Texas A & M Univ., had spent several months on a cruise in the Gulf of Mexico aboard Jacques Cousteau’s Calypso. He said they routinely staged shots, such as a “migration” of spiny lobsters which had just been pushed out a PVC pipe, and footage supposedly filmed in Antarctica, but in fact shot in the Gulf of Mexico in September, in 90 degree (F) heat – the crew wore shorts under their down-filled parkas. And J.C. himself (Cousteau, that is) was helicoptered out to the ship so he could appear in a few segments. And I just read about some well-publicized, and quite stunning, video footage of a human fetus in the womb – turns out the fetus was a wax model. Fortunately, as the Discovery Channel’s Planet Earth series shows, it is possible to depict nature as it really is, if you’re willing to spend a lot of money and a heck of a lot of time, and put the photographers at great risk, to get the shots.

  38. 38
    Bruce Tabor says:

    Re. 11 Gavin said, “…physical modelling is the much better way to go (obviously since I work in a GCM group). However, I disagree that this cannot be tested. The models that are built can be tested in dozens of ways to verify key feedbacks and mechanisms – response to volcanic eruptions, solar forcing, orbital forcing, ENSO variability, NAO or SAM responses, the last glacial maximum etc. And frankly, the projections from simulations done 20 years ago stand up very well to what actually happened.”

    This raises the complex issue of the link between types of experiment and quality of evidence.

    True, “correlation does not equal causation”, but we rely on this equality for randomised controled trials (i.e. experiments). Controlled experiments are generally regarded as providing the best level of evidence linking cause and effect. One variable is manipulated in the test group while held constant in the control group. Ideally all other variables are held constant or experimentally units are randomly allocated to the two groups so that statistical tests of outcome are valid. The difference in outcome between the test and control groups is observed. This experimental process also underlies the discovery of the physical laws programmed into GCMs.

    As for physical models, George Box said, “All models are wrong, some models are useful.”

    A classic recent example in supernova SN2006gy, the largest supernova ever observed, which fell outside the standard models of supernovae explosions (and provided physical evidence for an alternative model, the “pair instability model”):

    A recent failure of modelling bearing on GCMs may be that ice sheets are melting much faster than predicted – due to poorly understood mechanisms. (Ice sheets don’t just melt from the top.)

    Like astronomical models, climate models are not susceptible to normal experimental verification – test vs control. However, as Gavin opoints out, they make predictions and can be checked against those predictions. Together with environmental evidence, GCMs provide the best evidence of climate change in response to GHGs. It is possible that poorly understood mechanisms could invalidate aspects of the link between GHGs and temperature in GCMs, but the error could go either way. So only a fool would bet on GCMs being wrong. And it would be immoral for such a fool to make decisions over energy policy etc. on behalf of the rest of us.

  39. 39
    Svet says:

    On the topic of confusing data – can anyone tell me why the two global temperature graphs at
    GISS and the Hadley Centre (shown below) look so different for the last six years? The Hadley Centre graph appears to show a leveling off but the GISS graph does not. Is there some subtle difference in what they are measuring? If so, can anyone refer me to a resource that explains the differences?

    [Response: There is a difference in how the interpolate between data stations, particularly in the Arctic – HadCRU does not estimate Arctic ocean temperatures from nearby coastal data, while the GISS analysis does – given the warmth of the Arctic in recent years, that gives make the GISS anomalies slightly warmer. The ongoing sea ice retreat is probably corroborating evidence that this is a reasonable procedure. Look at the spatial maps of anomalies to see this more clearly. -gavin]

  40. 40

    I am a believer. Actually, I have been convinced by the aliens who took me from my island into their spacecraft : the Republicans are definitely a species which evoved from planet Neptune, where World Climate has found their best correlation proof.

  41. 41
    PaulM says:

    Thanks Gavin for pointing out that just because C02 has risen
    and temperatures have risen, it does not follow that one causes the other.

    As an alternative to the pirate theory, there is also the skirt hem height
    theory, levels of which were very low in the early 20th century but have risen
    alarmingly in recent years, correlating well with increased temperatures!
    Several possible mechanisms suggest themselves, such as increased metabolic
    rates of observers.
    This would also explain why warming has not occurred in Antarctica.

    As for Friis-Christensen and Svensmark, suggesting that the Sun might have
    something to do with the Earth’s climate, well, how ridiculous.
    Next they will be suggesting that the Pope is catholic.

    [Response: You misunderstand my point. I have no problem with the sun affecting the Earth’s climate – I have been an author on multiple papers demonstrating this (most recently Shindell et al, 2006) – and nowhere have I ever claimed that CO2 is important based solely on a correlation. FC&S 91 was bad not for what it purported to conclude, but for how it did it. That was evident at the time, and subsequent data showed that the relationship they found had no predictive quality. – gavin]

  42. 42
    Julian Flood says:

    Hmmm. I can do you a theory which ties mineral oil production to isotope signals — I don’t think the conventional CO2 theory does this very well as isotope signals appear in 1850 and AFAICT the real effect of greenhouse warming doesn’t show up until later. My theory even covers the SST temperature spike during WWII when the oil spill in the NH was enormous, mostly because of the Kreigesmarine submarine offensive. This inadvertent experiment shows a perfect rise in accordance with my oil sheen hypothesis.

    You now go and look at the Hadcru graphs and come back with the reasonable point that the SSTs begin to rise two years earlier than the submarine offensive. Aha, I riposte, you need to use the unadjusted graph: the graphs normally available are adjusted for a change in how the temperatures were taken, by bucket or in the engine intake. If you eliminate the adjustment then my theory is triumphantly vindicated. Intrigued by this, I wondered about the correction: apparently, without an SST correction, the GCMs don’t produce a realistic land temperature during the 30s and 40s. This is a great shame. In order to make the world fit my theory then I have to change back data which was altered to fit a rather more expensive theory. Perhaps my horse is at rather long odds in this race, not being the result of science crossed with supercomputer technology, but isn’t it a pity that the data was altered in the first place? Back to the drawing board.

    While on the subject of isotopes, is there a graph anywhere which covers a couple of hundred years of delta C13 values? I’ve looked round the web and can’t find anything. It needs to be a graph because I am a nurseryman of little brain and I need to look at pictures. I reckon I might be able to correlate leachate composition of volcanic ejecta and delta C13 values. Now that would be a really good bit of crackpot science!

    The whole article shows, of course, how important it is to use raw data and not to quibble when things go wrong. That’s why the oil sheen theory was born, an object lesson in the error of post hoc ergo propter hoc. Using raw data I think it’s still on all four legs. Just.


  43. 43
    nick says:


    I’m quite aware of what the proposed causes are for anthropogenic C02 affecting climate.

    However, the mistake you have made is this.

    You have made your plausible explanation the null hypothesis and not applied a test that shows that your plausible explanation is what actually has taken place.

    The null hypothesis is that there is no AGW.

    Part of the proof of AGW is to show that current climate diverges from the null hypothesis by a statistically significant amount. That has not been done at all.

    If it had been done, then the IPCC would be able to point out just how much of climate is affected by particular forcings. It can’t yet do that.

    [Response: You are incorrect. This cannot be given exactly due to the uncertainties in the forcing history (particular for aerosols), small variations in efficacy, and the remaining uncertainty in climate sensitivity. However, you can easily compare the impacts of different forcings – see here or here. -gavin]

    Let me explain more about the null hypothesis and a statistical test.

    The null hypothesis is no AGW. So we take climate prior to 1900 (industrialisation) and take the historical record for climate. Here we are talking about temperature proxies, orbital mechanics, solar proxies, vocanic eruptions, C02 records.

    Given this data it is possible to build an analysis that shows if a particular input affects the output, the extent to which it effects the output. At the end should be a random residual. If it is not random, then there is a missing input into the system.

    The systems will clearly be guided by the sorts of models of which you are aware. For example, its pretty clear how volcanos affect climate. We have lots of data to make an accurate model.

    Once you have a good model for the long term climate, you know look at the data post 1900.

    You can ask the question, do we need to introduce any extra variables such as anthropogenic C02 to get a fit?

    You can also ask, just how much of a difference does this extra C02 make and is it statistically significant.

    The IPCC can’t yet do this.

    Does it matter?

    Of course it matters. If anthropogenic C02 only makes up a small fraction of the change in climate, then a lot of money is going to be spent with very little resulting effect.


  44. 44
    David says:

    Huh. Your theory doesn’t explain Bernie Sanders. Because of that, I reserve the right to dismiss everything you say about anything, ever.

  45. 45
    Donald E. Flood says:

    I think that a distinction has to be made between prospective versus retrospective scientific studies. For instance, what is the probability that one could open a large phone book and find, at random, a page that contained five listings that ended in the same number? Answer: Quite low. However, start at the beginning of the phone book and search every page until the end, what then, is the probability of finding five listings that end in the same number? Answer: Virtually one.

    When was global warming identified as a major issue? Late 1970s? Certainly, by the mid-1980s, because the National Academy of Sciences was sponsoring educational programs on the subject! Are we to believe that it is mere coincidence, that is, “natural variability,” that the warmest years on record have occurred since then? Improbable, to say the least!

  46. 46
    CobblyWorlds says:

    #35, Svet,

    Vose et al; “An intercomparison of trends in surface air temperature analyses at the global, hemispheric, and grid-box scale.” GEOPHYSICAL RESEARCH LETTERS, VOL. 32, L18718, doi:10.1029/2005GL023502, 2005

    They find that when you use the same method of processing you get the same results for GHCN/GISS/CRU.

    The reduction of trend is because the Southern Hemisphere is warming less rapidly than the Southern Hemisphere, and CRU (on whom the Met Office graph is based) average the 2 hemispheres – that causes the difference. GISS use grid-box averaging. The apparent reduction of trend appears to be an artefact of processing.

    #28, Walter Manny,

    Your friend was right, this is an excellent site. If you’ve actually read the papers Gavin mentions the point of his seemingly frivolous article becomes all too apparent. There is a very serious point behind it about some dubious “science”.

    Thanks for another excellent post.

  47. 47
    Jens Olaf Pepke Pedersen says:

    I always found that RealClimate was just a political propaganda blog from a small group of climate politicians, so I am happy to note that there are actually real scientific discussions on your site.

  48. 48
    SomeBeans says:

    #35 – Svet

    The GISS and Hadley Centre datasets that you point to terminate at different years – Hadley runs to 2006, whilst GISS runs to 2003.

  49. 49
    Francis says:

    The truth of the republican reign has now been revealed. It is our duty to improve the world, even if all republicans must be tagged and sent to Australia.

  50. 50
    Dan Hughes says:

    More fun with correlations; some have even appeared in the peer-reviewed literature.